the demography of native and non-native plant species in

RE

VIS

ED

PRO

OF

ORIGINAL ARTICLE1

2 The demography of native and non-native plant species3 in mountain systems: examples in the Greater Yellowstone4 Ecosystem

5 Fredric W. Pollnac • Bruce D. Maxwell •

6 Mark L. Taper • Lisa J. Rew

7 Received: 28 November 2012 / Accepted: 2 July 20138 � The Society of Population Ecology and Springer Japan 2013

9 Abstract In mountainous areas, native and non-native

10 plants will be exposed to climate change and increased

11 disturbance in the future. Non-native plants may be more

12 successful than natives in disturbed areas and thus be able

13 to respond quicker to shifting climatic zones. In 2009,

14 monitoring plots were established for populations of a non-

15 native species (Linaria dalmatica) and a closely related

16 native species (Castilleja miniata) on an elevation gradient

17 in the Greater Yellowstone Ecosystem, USA. Population

18 data were collected twice during the growing season for

19 3 years and used to calculate population vital rates for both

20 species, and to construct population dynamics models for

21 L. dalmatica. Linaria dalmatica vital rates were more

22 associated with climatic/environmental factors than those

23 of C. miniata. Population dynamics models for L. dalm-

24 atica showed no trend in population growth rate (k) vs.

25 elevation. The highest k corresponded with the lowest

26 vegetation and litter cover, and the highest bare ground

27 cover. All populations with k\ 1 corresponded with the

28 lowest measured winter minimum temperature. There was

29 a negative association between k and number of weeks of

30 adequate soil moisture, and a weak positive association

31 between k and mean winter minimum temperature.

32Variance in vital rates and k of L. dalmatica suggest broad

33adaptation within its current range, with the potential to

34spread further with or without future changes in climate.

35There is evidence that k is negatively affected by persistent

36soil moisture which promotes the growth of other plant

37species, suggesting that it might expand further if other

38species were removed by disturbance.

39

40Keywords Climate change � Elevation gradient �41Invasive species � Linaria dalmatica �42Population model � Vital rates

43Introduction

44Plant communities in mountainous areas of the world are

45facing an uncertain future. Climate change has the potential

46to alter both native (Crimmins et al. 2009; Engler et al.

472009) and non-native (Becker et al. 2005; McDougall et al.

482005; Marini et al. 2009; Pauchard et al. 2009) plant spe-

49cies ranges. The broad geographic ranges and climatic

50tolerances of non-native plant species, as well as charac-

51teristics that may facilitate rapid range shifts in the face of

52climate change (Hellmann et al. 2008), could result in

53increased success of non-native plants at higher elevations

54(McDougall et al. 2005; Crimmins et al. 2009). In addition,

55increased and altered human land use in mountainous areas

56could result in more opportunities for non-native plant

57species establishment due to both an increase in dispersal

58vectors and in suitable habitat (McDougall et al. 2009;

59Pauchard et al. 2009).

60In general, non-native plant species richness tends to

61decrease at higher elevations in most geographical contexts

62(Pauchard et al. 2009; Alexander et al. 2011; Seipel et al.

632012). This response has been observed in some cases to be

A1 F. W. Pollnac (&) � B. D. Maxwell � L. J. Rew

A2 Department of Land Resources and Environmental Sciences,

A3 Montana State University, Leon Johnson Hall, Bozeman,

A4 MT 59717, USA

A5 e-mail: [email protected]

A6 M. L. Taper

A7 Department of Ecology, Montana State University,

A8 Bozeman, MT, USA

A9 M. L. Taper

A10 Department of Biology, University of Florida,

A11 Gainesville, FL, USA

123

Popul Ecol

DOI 10.1007/s10144-013-0391-4

Journal : Large 10144 Dispatch : 24-7-2013 Pages : 15

Article No. : 391h LE h TYPESET

MS Code : h CP h DISK4 4

RE

VIS

ED

PRO

OF

64 linear (Becker et al. 2005; McDougall et al. 2005), and to

65 be hump shaped in others (Tassin and Riviere 2003; Are-

66 valo et al. 2005). One recent exception to this generality

67 was found by Paiaro et al. (2011), who observed that non-

68 native species richness increased at both ends of an ele-

69 vation gradient in central Argentina. When considering

70 individual non-native species, several studies have noted

71 decreases in occurrence and population size and/or vital

72 rates with elevation (Liang et al. 2008; Alexander et al.

73 2009; Monty and Mahy 2009; Trtikova et al. 2010; Pollnac

74 et al. 2012) and at least one has noted increased success at

75 higher elevations (Ansari and Daehler 2010).

76 The presence of some exceptions to the general trends

77 suggests that elevation is not the only explanatory factor

78 related to non-native plant species distributions in moun-

79 tain systems, and it is clear that there are several other

80 factors besides elevation (i.e., climatic and environmental

81 characteristics) which influence non-native plant species in

82 these areas. Furthermore, the effects of these factors on any

83 given species could manifest themselves in subtle ways.

84 For example, imagine that seed viability is the only vari-

85 able negatively affected by climatic and/or environmental

86 factors for a particular species at high elevations, and all

87 other variables (such as stem density) remain constant at

88 high elevations. If seed viability is not measured, an

89 important relationship could go unnoticed and some

90 incorrect conclusions could be drawn based only on mea-

91 surements of stem density. Thus, if questions related to the

92 future of a non-native species in a mountainous area are to

93 be addressed, population level demographic details of the

94 species must be collected throughout its current range.

95 However, we know of no demographic studies for indi-

96 vidual non-native species which incorporate site-specific

97 measurements of climate and environmental factors.

98 The differences in the population growth rate (k) of a

99 plant species between its range limits and its interior range

100 have been hypothesized to vary based on whether or not

101 that species has reached the limits of its potential range

102 (Gaston 2003; Angert 2006; Eckhart et al. 2011). For

103 example, if a species has filled its potential range, k at the

104 margins should be lower than k for the interior populations

105 because the marginal populations have theoretically

106 encountered some limiting factor which suppresses growth

107 and prevents them from expanding further (Gaston 2003,

108 and references therein). Alternatively, for a species that has

109 not yet reached the limits of its potential range, marginal

110 populations may exhibit higher k values because the spe-

111 cies is in suitable habitat where limiting intraspecific

112 density dependence is absent (Gaston 2003, and references

113 therein). Eckhart et al. (2011) take this concept a step

114 further, stating that: (1) covariation of k with environ-

115 mental factors known to influence k across a species range

116 suggests lack of adaptation as the primary limit to a

117species’ current range and, (2) lack of association between

118k and range position, and increases in k at range limits both

119implicate dispersal as the primary factor defining range

120limits. Therefore, comparisons of k for interior and range

121limit populations could provide insight into whether or not

122a non-native species has reached the limits of its potential

123range within an area, and what the barriers to expansion

124may be. In a mountainous area, evaluation of k for a spe-

125cies throughout its current range allows for an assessment

126of the potential for the species to become established in

127upslope or down slope habitats under current climate

128conditions. Adding site-specific climatic and environmen-

129tal data further allows one to hypothesize about possible

130changes in range due to changing climatic and environ-

131mental conditions.

132The first objective of this study was to examine the

133relationships between specific population vital rates and

134site-specific environmental and climatic factors for a short

135lived perennial non-native plant species, Linaria dalmatica

136(L.) Mill, along an elevation gradient in the Greater Yel-

137lowstone Ecosystem (GYE). Using this species gave us the

138opportunity to test some of the aforementioned assump-

139tions about k in relation to range limits in the context of a

140mountain system using a non-native species with a known

141date and location of introduction (1957, in the town of

142Mammoth, B. Maxwell, personal communication). We

143hypothesized that vital rates of L. dalmatica would vary

144along the studied gradient in relation to climatic and

145environmental predictor variables, indicating the potential

146for climate induced changes in population dynamics. Our

147second objective was to compare the trends in vital rates

148for L. dalmatica to those of a closely related perennial

149native species, Castilleja miniata (Douglas ex Hook.),

150found along a similar elevation gradient. We hypothesized

151that vital rates of C. miniata would not vary throughout its

152distribution as much as those of L. dalmatica since it has

153been present in the area for a much longer period of time.

154This hypothesis relied on two assumptions. The first was

155that the residence time of L. dalmatica on the landscape has

156not been sufficient for this species to go through enough

157colonization/extinction events to have a distribution pattern

158which is matched with its ideal habitat (e.g., it is still found

159in areas where it might not be able to persist in the long

160term). The second was that the environment of the study

161was not subject to any recent abrupt and heterogeneous

162disturbances which could result in variable success of

163either species regardless of the historical suitability of the

164environment. To the best of our knowledge, both of these

165assumptions are valid. We do not believe that recent/

166ongoing climate change violates the second assumption

167because it is imposed on a broad scale compared to the

168ranges of these two species within the study area. Our third

169objective was to model population growth of L. dalmatica

Popul Ecol

123Journal : Large 10144 Dispatch : 24-7-2013 Pages : 15



RE

VIS

ED

PRO

OF

170 throughout its current range in our study area. We

171 hypothesized that population growth rates would be lower

172 at the upper elevation range limit of this species, indicating

173 a potential climatic barrier.

174 Methods

175 Study area and site selection

176 This study was conducted along an elevation gradient

177 within the GYE in the vicinity of Gardiner, MT,

178 45�0106000N, 110�4203300W, 1,598 m elevation. Three roads

179 were chosen as elevation transects in the area. Each of

180 these transects proceeded from elevations of approximately

181 1,700 m near the bottom of the Yellowstone River Valley

182 in Gardiner, MT for variable distances to elevations just

183 short of the highest elevation extent for the specific tran-

184 sect, but represented the highest point of road access

185 (58 km–2,900 m for transect 1, 18 km–2,400 m for tran-

186 sect 2, and 15 km–2,200 m for transect 3). Transect 1 was

187 within Yellowstone National Park, and the other two were

188 just north of the park boundary on United States Forest

189 Service roads.

190 During July and August of 2008, the three elevation

191 transects were surveyed for the presence of L. dalmatica

192 and C. miniata, both hereafter referred to collectively as the

193 test species. During initial surveys, an effort was made to

194 identify every distinct population of the test species present

195 within 5 to approximately 200 m of the elevation transects

196 (roads) from their lowest elevations up to the highest extent

197 that was navigable in a vehicle. Linaria dalmatica and C.

198 miniata occurred from 1,700 to 2,300 m and from 2,100 to

199 2,800 m along the elevation gradient, respectively. During

200 early spring 2009, three study sites were established for

201 each species on each one of the three elevation transects

202 (Fig. 1). Sites were selected from the pool of surveyed sites

203 to represent a relatively even spread of elevations along

204 each elevation transect. The sites essentially represented

205 the low, medium, and high ranges for L. dalmatica and C.

206 miniata in their respective local elevation distributions.

207 Four 1 m2 monitoring plots were established randomly at

208 each one of these sites by throwing quadrats in areas where

209 the test species was present. For L. dalmatica, it was not

210 possible to select sites which had a uniform population size

211 across the elevation gradient, and quadrats were thus dis-

212 tributed across a variable number of populations (distinct

213 patches of the species separated by [10 m) within the

214 spatial extent of each site (2,000–4,000 m2). Populations

215 ranged in size from 5 to 530 m2 for five out of the nine

216 sites, with the remaining four sites containing larger pop-

217 ulations of 1,000–3,000 m2. All C. miniata sites contained

218 populations of the species which were \200 m2 in size.

219Measurement of climate and environmental variables

220Temperature was measured every hour at each site using

221one Lascar EL-USB 1 temperature data logger (tempera-

222ture logger) placed *0.5 m off the ground during the

223growing seasons (June to early September) of 2010 and

2242011, and the winters of 2009/10 and 2010/11. Soil

225moisture was measured weekly at each site during the

226growing seasons of 2010 and 2011 using three Delmhorst

227gypsum blocks installed at random locations throughout

228each site at a depth of 15 cm in the ground following the

229procedure described in Aho and Weaver (2008).

230Throughout the growing seasons of 2010 and 2011, pre-

231cipitation was measured weekly at each site using one rain

232gauge per site (Taylor Pro gauge).

233Biotic environmental variables (canopy cover, percent

234of bare ground, litter, and vegetation cover) were estimated

235by the same observer at all sites during the growing season

236of 2011 from six random locations within the site for each

237variable except canopy cover, which was estimated at 4

238random locations. Soil samples were taken to a depth of

23910 cm (approximately 285 cm3) from ten random locations

240at each site during the growing season of 2010 and ana-

241lyzed for abiotic environmental variables (pH, organic

242matter, total nitrogen, potassium, and plant available

243phosphorous content).

244Estimation of vital rates

245Vital rates were collected for the purposes of: (1) deter-

246mining the influences of climate and environmental factors

247on the vital rates for each species, (2) comparing the degree

248of association between vital rates and climatic/environ-

249mental variables among the test species, and (3) building

250demographic population models for L. dalmatica. Plots

251were monitored during early June and late August of each

252field season from 2009 to 2011. During each session, a

2531 m2 frame divided into 16 parts was placed over each plot.

254The location of each stem in the grid was then drawn on a

255piece of tracing paper, denoting seedlings, vegetative

256stems, or flowering stems with distinct symbols. Stems

257which were obviously arising from a common root crown

258were drawn to be touching on the mapping data sheet, such

259that individuals could be counted more precisely.

260To estimate vital rates from early spring to late summer,

261sheets from subsequent monitoring sessions were overlaid

262(for example August 2009 was placed over June 2009) for

263each plot, and the number of stems which had: (1) transi-

264tioned from vegetative to flowering, (2) stayed vegetative,

265(3) died, or (4) appeared since the previous period were

266counted, as were the number of individuals which had

267survived, died, or appeared since the previous period.

268Similarly, looking at the time period from August to the

Popul Ecol




RE

VIS

ED

PRO

OF

269 following June, the number of individuals that survived,

270 perished, or appeared since the previous period was

271 recorded. This yielded information on stem production,

272 spring individual survival, transition to flowering stem,

273 estimated vegetative ramet production (for L. dalmatica

274 only), and fall individual survival rates. So few seedlings

275 were observed that seedling survival could not be estimated

276 from these data.

277 Linaria dalmatica seed production was estimated by

278 counting the number of seed capsules present within each

279 plot at each site during late August of each field season.

280Seed production per seed capsule was estimated by sam-

281pling forty-five capsules from plants outside of the moni-

282toring plots at each site. Fifteen seed capsules were

283collected from each of the lowest, middle, and highest

284regions of the stems, with no more than one capsule being

285collected from a given stem. Capsules were then dried at

286constant temperature (43 �C) and seeds within were

287counted. To evaluate germination rates, four batches of 50

288seeds were collected per site and placed in a germination

289chamber at 15 �C alternating 12 h light/12 h dark and

290monitored weekly for 5 weeks. To evaluate seedling

Fig. 1 Map of study area

showing plot locations

Popul Ecol




RE

VIS

ED

PRO

OF

291 survival, 5 pots of 5 seedlings at the cotyledon stage and 5

292 pots containing ten seeds (both using seed from the loca-

293 tion of placement) were randomly placed at each site in late

294 May/early June of 2011 and then monitored until mid-

295 September. Pots were placed so that they were level with

296 the ground surface and were not given any supplemental

297 irrigation or fertilization.

298 Seed predation data were also collected at each site

299 using four randomly located 8 cm 9 8 cm plastic trays.

300 Each tray had 2 holes drilled in the bottom, one for

301 drainage and one for securing the tray to the ground using a

302 nail. Trays were placed such that they were flush with the

303 ground surface. They were filled with sand, and 50 L.

304 dalmatica seeds were scattered on the surface of the sand.

305 After 10 days, the remaining seeds were collected and

306 counted. These data were used to refine seed production

307 rates for each site in the population model.

308 Seed production, germination, longevity, predation, and

309 seedling survival were used in conjunction with the vital

310 rates described above to model the population dynamics of

311 L. dalmatica. The only rate used in the models which was

312 not based on field measurements was seedling competitive

313 ability. Since seedlings of L. dalmatica are weak compet-

314 itors with established vegetation (Gates and Robocker

315 1960; Robocker 1970), germinable seed crop was further

316 reduced in the model by multiplying by the mean propor-

317 tion of bare ground measured at each site, assuming ran-

318 dom dispersal and random seed decay. That is, if a seed

319 landed on existing vegetation, it was expected to die, but if

320 it landed on bare soil, it was subject to germination. This

321 assumption is supported by previous work (F.W. Pollnac

322 and L.J. Rew, unpublished data), in which the presence and

323 cover of L. dalmatica was found to be positively associated

324 with increased bare ground, suggesting that establishment

325 of this species is linked to increased bare ground. It is quite

326 possible that seed decay and predation could also be related

327 to the environment in which a seed lands, but we did not

328 have data to test this.

329 For C. miniata, an estimate of seed production was

330 achieved by randomly harvesting 10 flowering heads from

331 outside of the monitoring plots at each of the C. miniata

332 sites in the late summer of 2009. These flowering heads

333 were dried at constant temperature (43 �C) for several

334 weeks and then dissected such that for each seed capsule,

335 the number of seeds could be counted. Seed germination

336 and seed decay rates for this species were not measured.

337 Data analysis

338 Variance in vital rates along elevation gradients

339 All analyses were performed using R 2.14.1 (R-Develop-

340 ment-Core-Team 2011). Before producing any population

341dynamics models, we wished to see if vital rates varied

342from site to site for each test species, and to see if there was

343any difference between species in the degree to which vital

344rates were associated with climatic/environmental vari-

345ables. To address these objectives, vital rate data collected

346directly from study plots [spring and fall individual sur-

347vival, stem production, transition to flowering stem, veg-

348etative ramet production (for L. dalmatica only), and seed

349production] from different years were averaged for each

3501 m2 plot, such that in the analysis, each plot at each site

351had one vital rate value, yielding a sample size of 4 per site

352for each vital rate for each species. Climate data collected

353in successive years (Table 1) were averaged for each site

354because the overarching interest was in how the trajectory

355of population growth might be affected by changes in vital

356rates as influenced by climatic factors over time. Although

357yearly fluctuations in climatic variables likely produce

358fluctuations in vital rates, the time lag between the two is

359unknown, and population growth over time is a product of

360the averages of such fluctuations.

361We employed a bootstrapped stepwise model selection

362procedure for analyzing these data to avoid pseudo-repli-

363cation imposed by the structure of our study. For each of

3641,000 bootstrap replicates, a random dataset was generated

365from our data by selecting 9 sites with replacement from

366our pool of sites. An information criterion approach

367(Burnham and Anderson 2002) was then used to determine

368which set of environmental and climatic predictor variables

369best explained the variance in each of the vital rates ana-

370lyzed for each iteration. Full models contained elevation

371and all of the environmental and climatic predictor vari-

372ables (Table 1) and any second order polynomial terms

373deemed necessary by examination of diagnostic plots of

374each vital rate plotted against individual predictor vari-

375ables. A stepwise model selection procedure with both

376backward and forward selection was applied to the full

377model, and the resulting best model (based on Akaike’s

378Information Criterion, AIC) for each iteration was recor-

379ded. Bar graphs depicting the number of best models

380containing different numbers of environmental/climatic

381variables were then qualitatively examined to compare the

382degree of association between vital rates and environ-

383mental/climatic variables for each species. Cases in which

384models without environmental/climatic variables were

385selected as the best model in 75 % of iterations were

386treated as evidence that these variables had little influence

387on that particular vital rate.

388Population growth rate of Linaria dalmatica

389along elevation gradients

390Population dynamics were modeled for each site using a

391difference equation model (to accommodate the two

Popul Ecol




RE

VIS

ED

PRO

OF

392 transitions) which was based on the vital rate data collected

393 from each of the four individual plots. At each site, k was

394 estimated from vital rates in each of 3 years for each of the

395 four plots. We wished to compare k among sites accounting

396 as best we could for both temporal variation among years

397 and spatial variation among plots. We characterized the

398 overall k at each site as the arithmetic mean over plots of

399 the geometric mean k for each plot over years. We assessed

400 the uncertainty in this measure with a parametric bootstrap

401 as follows. For each site, we fit a linear mixed model

402 (function lmer in package lme4 in R) to the 12 log trans-

403 formed ks with the overall mean as a fixed effect, and

404 variance components for year, plot, and residual error. For

405 each of 1,000 bootstrap replicates, we generated 12

406 observed log ks by combining the site mean with 3 years

407 effects, 4 plot effects, and 12 residual errors each drawn

408 from centered normal distributions using the appropriate

409 variances. The log ks were back transformed to ks and the

410 arithmetic mean of the geometric mean ks was calculated

411 and stored.

412 Separate mixed effects models were used to quantify the

413 trend in k in response to the fixed effects of elevation and

414 each individual climatic or environmental variable, using

415 the k values from the parametric bootstrap. Site was

416included as a random effect to account for the temporal

417pseudo-replication inherent in the simulation and the spatial

418pseudo-replication imposed by study design. In addition,

419median k values were assessed qualitatively for differences

420based on site characteristics using box and whisker plots.

421Although this qualitative assessment cannot generate any

422statements of statistical significance, we believe it is of

423value for its potential to elucidate trends which may be

424biologically significant which will aid in generating new

425hypotheses to test more quantitatively in the future. We

426looked for obvious shifts from k [ 1 to k \ 1 based on

427environmental characteristics. Instances where the highest

428or lowest median value of k was positioned at either end of

429the range of the climatic or environmental variable being

430examined were viewed as indications that extreme values of

431k may be related to extreme levels of the variable.

432Results

433Linaria dalmatica and Castilleja miniata vital rates

434Linaria dalmatica sites were found to be variable in terms

435of both climate and environmental conditions (Table 1).

Table 1 Environmental and climate characteristics of the nine L. dalmatica study sites

Site 1_1 1_2 1_3 2_1 2_2 2_3 3_1 3_2 3_3

Bare ground (%) 54.5 25.3 67.0 32.2 47.3 37.5 80.7 38.3 38.0

Canopy closure (%) 0.0 8.0 3.0 0.0 23.8 30.9 0.0 0.0 18.7

Elevation (m) 1737 2002 2237 1876 2015 2318 1785 1875 2159

Growing degree (days)* 1117.1 572.3 888.5 NA 960.6 781.4 1224.1 NA 895.8

Gs. frost free (days) 91 74 89 89 91 86 91 90 87

Gs. mean min. temperature (�C) 7.30 2.76 6.46 6.05 7.07 5.21 8.16 6.16 5.23

Gs. min. temperature (�C) -1.8 -4.8 -1.8 -3.0 -1.5 -3.0 -0.8 -2.5 -3.3

Litter cover (%) 16.7 19.8 8.0 30.5 12.7 6.2 3.0 24.0 20.2

Soil pH 7.1 6.5 6.9 7.1 7.2 6.6 7.1 6.7 6.7

Precipitation (cm) 6.1 7.0 5.1 6.1 7.4 10.5 3.7 6.2 7.5

Total soil nitrogen (ppm) 1.5 4.5 2.5 2.0 1.5 5.5 1.5 3.0 1.5

Soil organic matter (%) 3.6 9.3 4.3 4.9 2.4 4.2 3.2 7.7 5.3

Soil phosphorous (ppm) 13.0 29.0 27.0 19.0 10.0 27.0 13.0 29.0 26.0

Soil potassium (ppm) 534.0 560.0 323.0 553.0 419.0 442.0 478.0 533.0 340.0

Vegetation cover (%) 28.8 54.8 26.8 37.3 40.0 56.3 16.5 37.7 41.8

Weeks of ad. soil moist. 6.0 7.0 7.5 6.0 6.5 7.0 4.0 6.0 7.0

Wint. frost (days) 211.5 251.0 224.0 NA 217.5 168.5 206.0 NA 230.5

Wint. mean min. temperature (�C) -5.1 -6.4 -6.4 NA -4.3 -3.3 -4.4 NA -4.0

Wint. min. temperature (�C) -29.3 -29.3 -29.3 NA -25.3 -20.3 -26.8 NA -23.8

For site, the first number is the transect identifier, and the second number is the site identifier (1 low elevation, 2 middle elevation, 3 high

elevation)

Gs growing season, Wint. winter, min. temperature minimum temperature, ad. soil moist adequate soil moisture (C-1.5 MPa), NA not available

due to failure of data logger

* Calculated with base 10 �C

Popul Ecol




RE

VIS

ED

PRO

OF

436 Vital rates also qualitatively appeared to vary from site to

437 site (Fig. 2). Castilleja miniata sites also showed vari-

438 ability in climate and environmental conditions (Table 2),

439 but vital rates appeared to be less variable in relation to

440 both climate and environmental conditions than those of L.

441 dalmatica (Fig. 3). For all of the measured C. miniata vital

442 rates except seed production, the majority of best models

443 from the boot-strapped analysis did not contain any cli-

444 matic/environmental variables (Fig. 4). For L. dalmatica,

445 the majority of best models included environmental or

446 climatic variables for most vital rates, except the spring

447 survival and seed production vital rates (Fig. 5). The fre-

448 quency with which different climatic/environmental vari-

449 ables were included in best models for each species is

450 displayed in Table 3.

451 Population growth rate of Linaria dalmatica

452 The mean projected population growth rate (k) for L.

453 dalmatica was found to be variable between sites (Fig. 6)

454 but there was no clear trend in k along the elevation gra-

455 dient (Fig. 7). Based on the mixed effects models, there

456 was a negative association between number of weeks of

457 adequate soil moisture (C-1.5 Mpa) and k (P = 0.04), and

458 the suggestion of a positive association between winter

459mean minimum temperature and k (P = 0.07). From the

460box plots, there were very few instances in which the

461distribution of k showed any consistent pattern across

462levels of the climatic and environmental variables mea-

463sured at each site. However, some weak patterns were

464noted. Lambda appeared to decrease as the number of

465weeks adequate soil moisture increased (Fig. 7), and to

466increase with increasing winter mean minimum tempera-

467ture (Fig. 8). Winter minimum temperature appeared to

468have some influence on k, in that the only sites where k\1

469had the lowest measured minimum temperature (-29.5 �C,

470Fig. 8). Lambda did not show any notable relationships

471with any soil characteristics (Fig. 9). The highest value of k472was observed at the lowest value of vegetation cover

473(excluding L. dalmatica), the lowest value of percent litter

474cover, and the highest value of percent bare ground

475(Fig. 10).

476Discussion

477Associations between vital rates and the environment

478Several of the vital rates for L. dalmatica were influenced

479by climate and environmental predictors, as has been

Fig. 2 Mean transition rates for each L. dalmatica site. Error bars represent 95 % confidence interval for the mean. For site ID, the first number

is the elevation transect identifier, and the second number is the site identifier (1 low elevation, 2 mid elevation, 3 high elevation). n = 36

Popul Ecol




RE

VIS

ED

PRO

OF

Table 2 Environmental and climate characteristics of the nine C. miniata study sites

Site 1_1 1_2 1_3 2_1 2_2 2_3 3_1 3_2 3_3

Bare ground (%) 0.0 21.5 29.8 1.5 2.3 5.0 0.3 0.0 3.5

Canopy closure (%) 39.4 26.3 33.2 47.6 35.4 12.0 43.8 77.4 60.5

Elevation (m) 2096 2683 2812 2151 2303 2368 2159 2183 2239

Growing degree (days)* NA 432.3 337.0 667.8 769.9 507.5 913.0 723.4 NA

Gs. frost free (days) NA 76 63 85 83 81 88 86 NA

Gs. mean min. temperature (�C) NA 4.7 4.1 4.3 4.8 3.9 5.5 4.2 NA

Gs. min. temperature (�C) NA -3.5 -2.5 -3.5 -3.5 -3.3 -2.8 -2.5 NA

Litter cover (%) 27.2 29.2 46.0 16.8 23.2 21.2 21.8 32.5 26.8

pH 6.3 6.0 6.0 6.9 6.2 6.4 6.3 6.4 5.8

Precipitation (cm) NA NA NA 9.3 11.3 15.7 7.7 6.6 10.2

Soil nitrogen (ppm) 1.0 3.5 4.5 2.5 7.5 5.0 9.5 0.5 1.5

Soil organic matter (%) 9.3 6.6 4.0 6.2 9.0 7.3 11.4 7.2 27.2

Soil phosphorous (ppm) 16.0 52.0 30.0 4.0 17.0 33.0 29.0 40.0 18.0

Soil potassium (ppm) 258.0 603.0 316.0 238.0 400.0 358.0 471.0 371.0 372.0

Vegetation cover (%) 95.4 51.8 25.8 89.6 84.1 85.1 86.2 77.1 72.3

Weeks of ad. soil moist. 8.0 9.0 9.5 9.5 8.5 12.0 9.5 11.0 12.0

Wint. frost (days) 256.5 246.0 243.5 242.0 238.5 245.0 227.0 206.0 226.0

Wint. mean min. temperature (�C) -8.6 -4.2 -3.0 -7.5 -3.8 -6.8 -1.4 -3.2 -1.8

Wint. min. temperature (�C) -33.5 -29.0 -21.3 -30.0 -22.0 -30.0 -22.0 -19.0 -17.5

For site, the first number is the transect identifier, and the second number is the site identifier (1 low elevation, 2 middle elevation, 3 high elevation)

Gs growing season, Wint. winter, min. temperature minimum temperature, ad. soil moist adequate soil moisture (C-1.5 MPa), NA not available dueto failure of data logger

* Calculated with base 10 �C

Fig. 3 Mean vital rates for each C. miniata site. Error bars represent 95 % confidence interval for the mean. For site ID, the first number is the

transect identifier, and the second number is the site identifier (1 low elevation, 2 mid elevation, 3 high elevation). n = 36

Popul Ecol




RE

VIS

ED

PRO

OF

480 shown for other plant species along environmental gradi-

481 ents (Mack and Pyke 1984; Carlsson and Callaghan 1994;

482 Chambers et al. 2007; Purves 2009; Eckhart et al. 2011;

483 Gimenez-Benavides et al. 2011). As we had hypothesized,

484 L. dalmatica vital rates were associated with variation in

485 climatic and environmental conditions to a greater extent

486 than those of the closely related native species C. miniata.

487 This is most likely due to the difference in the level of

488 adaptation of each species to its environment (Eckhart et al.

489 2011 and references therein) based on the length of time

490 that each species has had to adapt to conditions within its

491 current range.

492 The fact that the vital rates for C. miniata did not seem

493 to be as influenced by climatic or environmental variables

494 is not surprising. Climatic conditions varied less during the

495 study within C. miniata’s current range than in the range of

496 L. dalmatica (F.W. Pollnac and L.J. Rew, unpublished

497 data). Notably, winter minimum temperatures were stable

498 within C. miniata’s range (F.W. Pollnac and L.J. Rew,

499 unpublished data). Thus, perhaps the lack of variability in

500 vital rates based on climate and environmental conditions

501 is due in part to the fact that these conditions were less

502 variable over the surveyed range for C. miniata than they

503were for L. dalmatica. This would suggest that C. miniata,

504having had more time to equilibrate within its range by

505going through colonization and extinction events, has

506occupied a geographic range where its vital rates are rel-

507atively stable due to more constant climatic conditions. It is

508also possible that C. miniata has had time to adapt to the

509variability present within its current range such that its vital

510rates can remain stable in spite of climatic variation, as has

511been hypothesized by Eckhart et al. (2011). Although we

512cannot formally test either of these hypotheses, our data

513suggest that it is a combination of both, given that there

514was less variability in most (but not all) of the climatic

515conditions along this species’ elevation range (F.W. Poll-

516nac and L.J. Rew, unpublished data), and that whatever

517variability there was did not seem to affect the vital rates of

518the species to any great extent. In contrast, L. dalmatica is a

519relative newcomer to the area, and therefore its vital rates

520may be more susceptible to climatic/environmental varia-

521tions because it has not had time to either adapt or go

522extinct in marginal environments where its vital rates may

523be adversely affected. This generally reflects the concept of

524the taxon cycle which states that species with longer resi-

525dence times tend to exhibit contracted ranges in interior

Fig. 4 Number of cases from a

1000 iteration boot-strap

procedure where vital rate best

models from stepwise AIC

model selection contained

different numbers of

environmental and climatic

variables for Castilleja miniata

Popul Ecol




RE

VIS

ED

PRO

OF

Table 3 Frequency with which the listed variables (top row) were included in best models for each dependent variable

Dependent

variable

Gs. mean min.

temperature (�C)

Polynominal Gs.

mean min. temperature (�C)

Wint. mean min.

temperature (�C)

Gs. frost free

(days)

Wint. frost

(days)

L. dalmatica Stem production 0.33 0.00 0.08 0.04 0.00

Spring survival 0.42 0.00 0.23 0.09 0.00

Fall survival 0.74 0.00 0.50 0.25 0.03

Vegetative reproduction 0.71 0.00 0.45 0.27 0.03

Seed production 0.40 0.03 0.11 0.00 0.00

Transition to flowering 0.47 0.44 0.22 0.00 0.00

C. Miniata Stem production 0.06 0.00 0.00 0.00 0.00

Spring survival 0.07 0.00 0.06 0.04 0.00

Fall survival 0.22 0.00 0.20 0.09 0.00

Seed production 0.54 0.00 0.37 0.02 0.00

Transition to flowering 0.08 0.00 0.10 0.07 0.00

Model selection was performed for 1000 bootstrapped replicates using a stepwise selection procedure with backward and forward selection based

on AIC

Gs growing season, Wint. winter, min. temperature minimum temperature

Fig. 5 Number of cases from a

1000 iteration boot-strap

procedure where vital rate best

models from stepwise AIC

model selection contained

different numbers of

environmental and climatic

variables for Linaria dalmatica

Popul Ecol




RE

VIS

ED

PRO

OF

526habitats whereas colonizing/ruderal species exhibit expan-

527ded ranges in marginal habitats (Wilson 1959). It also

528suggests that if climate were to change in the future, L.

529dalmatica’s range would be less likely to shrink (with its

530comparably broader tolerance to a variety of climatic

531conditions, including temperature) than would C.

532miniata’s.

533Population growth rate of Linaria dalmatica and its

534potential to spread to higher elevations

535There was no decrease in k with increased elevation as we

536had hypothesized. The lack of a decrease in k at the current

537high elevation limit of this species suggests that it may not

538yet have reached the limits of its potential range (Gaston

5392003). However, there was weak evidence that k for this

540species was influenced by some of the measured climate or

541environmental variables. In addition, although overall seed

542production for L. dalmatica increases with elevation, ger-

543mination rates decrease at the highest elevations (F.W.

544Pollnac and L.J. Rew, unpublished data). These results

Fig. 6 Boxplots of the distribution of the projected growth rate (k)

values for L. dalmatica by site (site ID) from a parametrically

bootstrapped matrix, based on estimated plot and year variance

components. n = 1000 for each site. For site ID, the first number is

the transect identifier, and the second number is the site identifier (1

low elevation, 2 mid elevation, 3 high elevation)

Fig. 7 Boxplots of distribution of growth rate (k) values for L. dalmatica by elevation and growing season climate variables from a

parametrically bootstrapped matrix, based on estimated plot and year variance components

Popul Ecol




RE

VIS

ED

PRO

OF

545 suggest that: (1) dispersal (Eckhart et al. 2011) and/or

546 propagule pressure may not be the primary limit to this

547 species’ current range, and (2) there may still be some

548 climatic limit which is preventing the seeds of this species

549 from germinating and/or seedlings from establishing at

550 higher elevations.

551 The general lack of strong relationships between k and

552 many of the measured climate variables suggests that L.

553 dalmatica may be able to tolerate a broad range of climatic

554 conditions. Other studies have suggested that non-native

555 plants which successfully invade mountain systems must

556 be broadly adapted to cope with the variable climatic

557 conditions found along the elevation gradients in these

558 areas (Alexander et al. 2011). However, winter minimum

559 temperature had an interesting relationship with k, in that

560 the only sites where k \ 1 corresponded to the lowest

561 measured winter minimum temperature. This suggests that

562 this species may only be broadly adapted to a point, and

563 that success of this species above its current elevation

564range may be limited by extremely low winter tempera-

565tures, as is common with plants in cold environments

566(Stocklin and Baumler 1996; Hobbie and Chapin 1998).

567Anything that would tend to increase winter temperatures,

568be it insulation due to increased snow pack or increased air

569temperatures under a warming climate, may favor the over

570winter survival of this species. Given the sensitivity of

571population levels to the over-winter survival rate (data not

572shown), this could result in large increases in population

573size. We have hypothesized that the location of this sur-

574vival barrier could be shifted in the future based on

575increased snow pack prior to extremely cold temperature

576events and/or a warming climate. However, properties of

577the vegetative community also appear to be exerting

578influence on k of L. dalmatica throughout its current ele-

579vation range.

580Population growth appeared to increase with decreased

581number of weeks of adequate soil moisture. Those sites

582with more persistent soil moisture generally had higher

583levels of vegetative cover and lower levels of bare ground

584(Table 1). The fact that the highest value of k for this

585species occurred where both vegetation and litter cover

586were the lowest and where percent bare ground was the

587highest suggests that the relationship between k and weeks

588of adequate soil moisture is related to increased growth of

589other vegetation and consequent litter production. These

590patterns suggest that while the current range of establish-

591ment of this species may be limited by climate, established

592populations may be primarily limited by characteristics of

593the vegetative community. Robocker (1974) noted that this

594species has low competitive ability in established perennial

595communities. Other studies have also shown negative

596associations between single non-native species abundances

597and vegetative community characteristics such as native

598species richness (Knight and Reich 2005) or native species

599diversity (Ortega and Pearson 2005), and that increased

600plant litter can decrease establishment of non-native plants

601(Hager 2004; Bartuszevige et al. 2007). Our results follow

602the same pattern, which suggests that if areas within or just

603outside of L. dalmatica’s current range in the GYE were to

604become more disturbed, which increases bare ground, this

605species would be likely to expand its range and/or the

606extent of current populations as a result.

607In the absence of establishment limitations, the lack of

608any strong climate/environmentally induced trends in k609suggests that this species could potentially spread outside

610of its present range under current climatic conditions. The

611lack of a consistent decrease in k at the upper elevation

612limits of this species is further evidence of this. Addi-

613tionally, while germination of seeds from high elevation

Fig. 8 Boxplots of distribution of growth rate (k) values for L.

dalmatica by winter climate variables from a parametrically boot-

strapped matrix, based on estimated plot and year variance

components

Popul Ecol




RE

VIS

ED

PRO

OF

614 sources was lower, the fact that propagule pressure for this

615 species is not constrained at higher elevations suggests that

616 it could successfully spread upward in the absence or

617 reduction of climatic barriers (F.W. Pollnac and L.J. Rew,

618 unpublished data). It is still possible that climate may be

619 limiting the establishment of L. dalmatica above its current

620 elevation limits, but our current data do not provide con-

621 clusive evidence of this. In the future, more specific tests of

622 germination, establishment, and survival need to be con-

623 ducted to test the hypothesis that this species is currently

624 experiencing an establishment based climatic limit to fur-

625 ther spread to higher elevations.

626 Although established L. dalmatica plants are viewed as

627 competitive, in that increased L. dalmatica density has

628 been shown to be associated with decreased density of

629 other plants (Robocker 1974; Wilson et al. 2005), whether

630 or not this species is capable of displacing other vegetation

631 is still questionable. Seedlings were rare in this study, and

632 are noted to not be particularly competitive with estab-

633 lished vegetation (Gates and Robocker 1960; Robocker

6341974) so this species may have difficulty establishing in

635heavily vegetated areas. However, the alpine zone is sub-

636ject to frequent natural soil disturbances (e.g., frost heaving

637and animal burrowing), is relatively sparse in established

638vegetation, and is likely to experience increased anthro-

639pogenic disturbance in the future. Thus, in the absence of

640climate constraints, the alpine/subalpine zone would seem

641to be an ideal habitat for potential L. dalmatica establish-

642ment. Since L. dalmatica has been shown to be broadly

643adapted and we have not been able to provide any con-

644clusive evidence of climatic limitation for this species,

645populations at its upper range limits should not be ignored

646in management efforts. In addition, areas above its current

647elevation range should be surveyed frequently for the

648presence of this species in order to prevent the spread of

649this species into higher elevation environments. Due to the

650sensitive nature of alpine habitats and the large proportion

651of plant diversity and endemic species contained therein

652(Korner et al. 2011), the impacts of non-native plant spe-

653cies in these areas could be particularly harsh. This, in

Fig. 9 Boxplots of distribution of growth rate (k) values for L. dalmatica by environmental variables from a parametrically bootstrapped matrix,

based on estimated plot and year variance components

Popul Ecol




RE

VIS

ED

PRO

OF

654 itself, may be enough justification to increase efforts to

655 limit invasions of non-native plant species, such as L.

656 dalmatica, into these areas.

657 Acknowledgments We would like to thank the Strategic Environ-658 mental Research and Development Program (project RC 1545), NRI659 2009-55320-05033, and NSF (GK-12 Grant # 0440594) for providing660 funding for this project. We would also like to thank the United States661 Forest Service and the National Park Service for their cooperation.662 Thanks to the MIREN consortium and the MSU Weed and Invasive663 Plant Ecology and Management group for their input and support. We664 would like to thank 2 anonymous reviewers for their help in665 improving this manuscript, and Tyler Brummer for his assistance with666 R. Finally, we would also like to thank Adam, Barb, Alex, Kim,667 Landon, Curtiss, Jordan, and Mel C for assistance in the field.

668 References

669 Aho K, Weaver T (2008) Measuring soil water potential with gypsum670 blocks: calibration and sensitivity. Intermountain J Sci 14:51–60

671Alexander JM, Naylor B, Poll M, Edwards PJ, Dietz H (2009) Plant672invasions along mountain roads: the altitudinal amplitude of673alien asteraceae forbs in their native and introduced ranges.674Ecography 32:334–344675Alexander JM, Kueffer C, Daehler CC, Edwards PJ, Pauchard A,676Seipel T (2011) Assembly of nonnative floras along elevational677gradients explained by directional ecological filtering. Proc Natl678Acad Sci USA 108:656–661679Angert AL (2006) Demography of central and marginal populations680of monkeyflowers (Mimulus cardinalis and M. lewisii). Ecology68187:2014–2025682Ansari S, Daehler CC (2010) Life history variation in a temperate683plant invader, Verbascum thapsus along a tropical elevational684gradient in Hawai’i. Biol Invasions 12:4033–4047685Arevalo JR, Delgado JD, Otto R, Naranjo A, Salas M, Fernandez-686Palacios JM (2005) Distribution of alien vs. native plant species in687roadside communities along an altitudinal gradient in Tenerife and688Gran Canaria (Canary Islands). Perspect Plant Ecol 7:185–202689Bartuszevige AM, Hrenko RL, Gorchov DL (2007) Effects of leaf690litter on establishment, growth and survival of invasive plant691seedlings in a deciduous forest. Am Midl Nat 158:472–477692Becker T, Dietz H, Billeter R, Buschmann H, Edwards PJ (2005)693Altitudinal distribution of alien plant species in the Swiss Alps.694Perspect Plant Ecol 7:173–183695Burnham KP, Anderson DR (2002) Model selection and multimodel696inference: a practical information-theoretic approach. Springer,697New York698Carlsson BA, Callaghan TV (1994) Impact of climate-change factors699on the clonal sedge Carex bigelowii—implications for popula-700tion-growth and vegetative spread. Ecography 17:321–330701Chambers JC, Roundy BA, Blank RR, Meyer SE, Whittaker A (2007)702What makes great basin sagebrush ecosystems invasible by703Bromus tectorum? Ecol Monogr 77:117–145704Crimmins TM, Crimmins MA, Bertelsen CD (2009) Flowering range705changes across an elevation gradient in response to warming706summer temperatures. Global Change Biol 15:1141–1152707Eckhart VM, Geber MA, Morris WF, Fabio ES, Tiffin P, Moeller DA708(2011) The geography of demography: long-term demographic709studies and species distribution models reveal a species border710limited by adaptation. Am Nat 178:S26–S43711Engler R, Randin CF, Vittoz P, Czaka T, Beniston M, Zimmermann712NE, Guisan A (2009) Predicting future distributions of mountain713plants under climate change: does dispersal capacity matter?714Ecography 32:34–45715Gaston KJ (2003) The structure and dynamics of geographic ranges.716Oxford University Press, New York717Gates DH, Robocker WC (1960) Revegetation with adapted grasses718in competition with Dalmatian toadflax and St. Johnswort.719J Range Manage 13:322–326720Gimenez-Benavides L, Albert MJ, Iriondo JM, Escudero A (2011)721Demographic processes of upward range contraction in a long-722lived Mediterranean high mountain plant. Ecography 34:85–93723Hager HA (2004) Differential effects of typha litter and plants on724invasive Lythrum salicaria seedling survival and growth. Biol725Invasions 6:433–444726Hellmann JJ, Byers JE, Bierwagen BG, Dukes JS (2008) Five727potential consequences of climate change for invasive species.728Conserv Biol 22:534–543729Hobbie SE, Chapin FS (1998) An experimental test of limits to tree730establishment in Arctic tundra. J Ecol 86:449–461731Knight KS, Reich PB (2005) Opposite relationships between inva-732sibility and native species richness at patch versus landscape733scales. Oikos 109:81–88734Korner C, Paulsen J, Spehn E (2011) A definition of mountains and735their bioclimatic belts for global comparisons of biodiversity736data. Alpine Bot 121:73–78

Fig. 10 Boxplots of distribution of growth rate (k) values for L.

dalmatica by environmental variables from a parametrically boot-

strapped matrix, based on estimated plot and year variance

components

Popul Ecol




RE

VIS

ED

PRO

OF

737 Liang Y, Jian L, Zang SP, Wang SJ, Guo WH, Wang RQ (2008)738 Genetic diversity of the invasive plant Coreopsis grandiflora at739 different altitudes in Laoshan mountain, China. Can J Plant Sci740 88:831–837741 Mack RN, Pyke DA (1984) The demography of Bromus tectorum—742 the role of microclimate, grazing and disease. J Ecol 72:731–748743 Marini L, Gaston KJ, Prosser F, Hulme PE (2009) Contrasting744 response of native and alien plant species richness to environ-745 mental energy and human impact along alpine elevation746 gradients. Global Ecol Biogeogr 18:652–661747 McDougall KL, Morgan JW, Walsh NG, Williams RJ (2005) Plant748 invasions in treeless vegetation of the Australian Alps. Perspect749 Plant Ecol 7:159–171750 McDougall K, Haider S, Seipel T, Kueffer C, MIREN Consortium751 (2009) Spread of non-native plant species into mountains: now is752 the time to act. Mountain Forum Bulletin 9:23–25753 Monty A, Mahy G (2009) Clinal differentiation during invasion:754 Senecio inaequidens (Asteraceae) along altitudinal gradients in755 Europe. Oecologia 159:305–315756 Ortega YK, Pearson DE (2005) Weak vs. Strong invaders of natural757 plant communities: assessing invasibility and impact. Ecol Appl758 15:651–661759 Paiaro V, Cabido M, Pucheta E (2011) Altitudinal distribution of760 native and alien plant species in roadside communities from761 central Argentina. Austral Ecol 36:176–184762 Pauchard A, Kueffer C, Dietz H, Daehler CC, Alexander J, Edwards763 PJ, Arevalo JR, Cavieres LA, Guisan A, Haider S, Jakobs G,764 McDougall K, Millar CI, Naylor BJ, Parks CG, Rew LJ, Seipel T765 (2009) Ain’t no mountain high enough: plant invasions reaching766 new elevations. Front Ecol Environ 9:479–486767 Pollnac FW, Seipel T, Repath C, Rew LJ (2012) Plant invasion at768 landscape and local scales along roadways in the mountainous

769region of the Greater Yellowstone Ecosystem. Biol Invasions77014:1753–1763771Purves DW (2009) The demography of range boundaries versus range772cores in eastern US tree species. P R Soc B 276:1477–1484773R-Development-Core-Team (2011) R: a language and environment774for statistical computing. R Foundation for Statistical Comput-775ing, Vienna, Austria776Robocker WC (1970) Seed characteristics and seedling emergence of777Dalmatian toadflax. Weed Sci 18:720–725778Robocker WC (1974) Life history, ecology, and control of Dalmatian779toadflax. Technical Bulletin 79, Washington Agricultural Exper-780iment Station781Seipel T, Kueffer C, Rew LJ, Daehler CC, Pauchard A, Naylor BJ,782Alexander JM, Edwards PJ, Parks CG, Arevalo JR, Cavieres LA,783Dietz H, Jakobs G, McDougall K, Otto R, Walsh N (2012)784Processes at multiple scales affect richness and similarity of non-785native plant species in mountains around the world. Global Ecol786Biogeogr 21:236–246787Stocklin J, Baumler E (1996) Seed rain, seedling establishment and788clonal growth strategies on a glacier foreland. J Veg Sci 7:45–56789Tassin J, Riviere JN (2003) Species richness altitudinal gradient of790invasive plants on Reunion Island (Mascareigne Archipelago,791Indian Ocean). Revue D Ecol-Terre Vie 58:257–270792Trtikova M, Edwards PJ, Gusewell S (2010) No adaptation to altitude793in the invasive plant Erigeron annuus in the Swiss Alps.794Ecography 33:556–564795Wilson EO (1959) Adaptive shift and dispersal in a tropical ant fauna.796Evolution 13:122–144797Wilson LM, Sing SE, Piper GL, Hansen RW, De Clerck-Floate R,798MacKinnon DK, Randall C (2005) Biology and biological799control of Dalmation and Yellow toadflax. Report FHTET-2005-80013, USDA Forest Service

801

Popul Ecol




the demography of native and non-native plant species in

Documents